1 /*****************************************************************************
2 * cvd.c : CVD Subtitle decoder
3 *****************************************************************************
4 * Copyright (C) 2003, 2004 the VideoLAN team
7 * Authors: Rocky Bernstein
8 * Gildas Bazin <gbazin@videolan.org>
9 * Julio Sanchez Fernandez (http://subhandler.sourceforge.net)
10 * Laurent Aimar <fenrir@via.ecp.fr>
12 * This program is free software; you can redistribute it and/or modify
13 * it under the terms of the GNU General Public License as published by
14 * the Free Software Foundation; either version 2 of the License, or
15 * (at your option) any later version.
17 * This program is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU General Public License for more details.
22 * You should have received a copy of the GNU General Public License
23 * along with this program; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
25 *****************************************************************************/
27 /*****************************************************************************
29 *****************************************************************************/
34 #include <vlc_common.h>
35 #include <vlc_plugin.h>
37 #include <vlc_codec.h>
41 #define DEBUG_CVDSUB 1
43 /*****************************************************************************
45 *****************************************************************************/
46 static int DecoderOpen ( vlc_object_t * );
47 static int PacketizerOpen( vlc_object_t * );
48 static void DecoderClose ( vlc_object_t * );
51 set_description( N_("CVD subtitle decoder") )
52 set_capability( "decoder", 50 )
53 set_callbacks( DecoderOpen, DecoderClose )
56 set_description( N_("Chaoji VCD subtitle packetizer") )
57 set_capability( "packetizer", 50 )
58 set_callbacks( PacketizerOpen, DecoderClose )
61 /*****************************************************************************
63 *****************************************************************************/
64 static subpicture_t *Decode( decoder_t *, block_t ** );
65 static block_t *Packetize ( decoder_t *, block_t ** );
66 static block_t *Reassemble ( decoder_t *, block_t * );
67 static void ParseMetaInfo ( decoder_t *, block_t * );
68 static void ParseHeader ( decoder_t *, block_t * );
69 static subpicture_t *DecodePacket( decoder_t *, block_t * );
70 static void RenderImage( decoder_t *, block_t *, subpicture_region_t * );
72 #define SUBTITLE_BLOCK_EMPTY 0
73 #define SUBTITLE_BLOCK_PARTIAL 1
74 #define SUBTITLE_BLOCK_COMPLETE 2
80 int i_state; /* data-gathering state for this subtitle */
82 block_t *p_spu; /* Bytes of the packet. */
84 size_t i_spu_size; /* goal for subtitle_data_pos while gathering,
85 size of used subtitle_data later */
87 uint16_t i_image_offset; /* offset from subtitle_data to compressed
89 size_t i_image_length; /* size of the compressed image data */
90 size_t first_field_offset; /* offset of even raster lines */
91 size_t second_field_offset; /* offset of odd raster lines */
92 size_t metadata_offset; /* offset to data describing the image */
93 size_t metadata_length; /* length of metadata */
95 mtime_t i_duration; /* how long to display the image, 0 stands
96 for "until next subtitle" */
98 uint16_t i_x_start, i_y_start; /* position of top leftmost pixel of
99 image when displayed */
100 uint16_t i_width, i_height; /* dimensions in pixels of image */
102 uint8_t p_palette[4][4]; /* Palette of colors used in subtitle */
103 uint8_t p_palette_highlight[4][4];
106 /*****************************************************************************
107 * DecoderOpen: open/initialize the cvdsub decoder.
108 *****************************************************************************/
109 static int DecoderOpen( vlc_object_t *p_this )
111 decoder_t *p_dec = (decoder_t*)p_this;
112 decoder_sys_t *p_sys;
114 if( p_dec->fmt_in.i_codec != VLC_CODEC_CVD )
117 p_dec->p_sys = p_sys = malloc( sizeof( decoder_sys_t ) );
121 p_sys->b_packetizer = false;
123 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
126 p_dec->pf_decode_sub = Decode;
127 p_dec->pf_packetize = Packetize;
129 p_dec->fmt_out.i_cat = SPU_ES;
130 p_dec->fmt_out.i_codec = VLC_CODEC_YUVP;
135 /*****************************************************************************
136 * PacketizerOpen: open/initialize the cvdsub packetizer.
137 *****************************************************************************/
138 static int PacketizerOpen( vlc_object_t *p_this )
140 decoder_t *p_dec = (decoder_t*)p_this;
142 if( DecoderOpen( p_this ) != VLC_SUCCESS ) return VLC_EGENERIC;
144 p_dec->p_sys->b_packetizer = true;
149 /*****************************************************************************
150 * DecoderClose: closes the cvdsub decoder/packetizer.
151 *****************************************************************************/
152 void DecoderClose( vlc_object_t *p_this )
154 decoder_t *p_dec = (decoder_t*)p_this;
155 decoder_sys_t *p_sys = p_dec->p_sys;
157 if( p_sys->p_spu ) block_ChainRelease( p_sys->p_spu );
161 /*****************************************************************************
163 *****************************************************************************/
164 static subpicture_t *Decode( decoder_t *p_dec, block_t **pp_block )
166 block_t *p_block, *p_spu;
168 if( pp_block == NULL || *pp_block == NULL ) return NULL;
173 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
175 /* Parse and decode */
176 return DecodePacket( p_dec, p_spu );
179 /*****************************************************************************
181 *****************************************************************************/
182 static block_t *Packetize( decoder_t *p_dec, block_t **pp_block )
184 block_t *p_block, *p_spu;
186 if( pp_block == NULL || *pp_block == NULL ) return NULL;
191 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
193 p_spu->i_dts = p_spu->i_pts;
200 /*****************************************************************************
203 Data for single screen subtitle may come in several non-contiguous
204 packets of a stream. This routine is called when the next packet in
205 the stream comes in. The job of this routine is to parse the header,
206 if this is the beginning, and combine the packets into one complete
209 If everything is complete, we will return a block. Otherwise return
212 *****************************************************************************/
213 #define SPU_HEADER_LEN 1
215 static block_t *Reassemble( decoder_t *p_dec, block_t *p_block )
217 decoder_sys_t *p_sys = p_dec->p_sys;
220 if( p_block->i_buffer < SPU_HEADER_LEN )
222 msg_Dbg( p_dec, "invalid packet header (size %zu < %u)" ,
223 p_block->i_buffer, SPU_HEADER_LEN );
224 block_Release( p_block );
228 p_buffer = p_block->p_buffer;
230 /* From the scant data on the format, there is only only way known
231 * to detect the first packet in a subtitle. The first packet
232 * seems to have a valid PTS while later packets for the same
234 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY && p_block->i_pts == 0 )
236 msg_Warn( p_dec, "first packet expected but no PTS present");
240 p_block->p_buffer += SPU_HEADER_LEN;
241 p_block->i_buffer -= SPU_HEADER_LEN;
243 /* First packet in the subtitle block */
244 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY ) ParseHeader( p_dec, p_block );
246 block_ChainAppend( &p_sys->p_spu, p_block );
247 p_sys->p_spu = block_ChainGather( p_sys->p_spu );
249 if( p_sys->p_spu->i_buffer >= p_sys->i_spu_size )
251 block_t *p_spu = p_sys->p_spu;
253 if( p_spu->i_buffer != p_sys->i_spu_size )
255 msg_Warn( p_dec, "SPU packets size=%zu should be %zu",
256 p_spu->i_buffer, p_sys->i_spu_size );
259 msg_Dbg( p_dec, "subtitle packet complete, size=%zuu", p_spu->i_buffer);
261 ParseMetaInfo( p_dec, p_spu );
263 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
269 /* Not last block in subtitle, so wait for another. */
270 p_sys->i_state = SUBTITLE_BLOCK_PARTIAL;
277 We do not have information on the subtitle format used on CVD's
278 except the submux sample code and a couple of samples of dubious
279 origin. Thus, this is the result of reading some code whose
280 correctness is not known and some experimentation.
282 CVD subtitles are different in several ways from SVCD OGT subtitles.
283 Image comes first and metadata is at the end. So that the metadata
284 can be found easily, the subtitle packet starts with two bytes
285 (everything is big-endian again) that give the total size of the
286 subtitle data and the offset to the metadata - i.e. size of the
287 image data plus the four bytes at the beginning.
289 Image data comes interlaced is run-length encoded. Each field is a
290 four-bit nibble. Each nibble contains a two-bit repeat count and a
291 two-bit color number so that up to three pixels can be described in
292 four bits. The function of a 0 repeat count is unknown; it might be
293 used for RLE extension. However when the full nibble is zero, the
294 rest of the line is filled with the color value in the next nibble.
295 It is unknown what happens if the color value is greater than three.
296 The rest seems to use a 4-entries palette. It is not impossible
297 that the fill-line complete case above is not as described and the
298 zero repeat count means fill line. The sample code never produces
299 this, so it may be untested.
302 static void ParseHeader( decoder_t *p_dec, block_t *p_block )
304 decoder_sys_t *p_sys = p_dec->p_sys;
305 uint8_t *p = p_block->p_buffer;
307 p_sys->i_spu_size = (p[0] << 8) + p[1] + 4; p += 2;
309 /* FIXME: check data sanity */
310 p_sys->metadata_offset = (p[0] << 8) + p[1]; p +=2;
311 p_sys->metadata_length = p_sys->i_spu_size - p_sys->metadata_offset;
313 p_sys->i_image_offset = 4;
314 p_sys->i_image_length = p_sys->metadata_offset - p_sys->i_image_offset;
317 msg_Dbg( p_dec, "total size: %zu image size: %zu",
318 p_sys->i_spu_size, p_sys->i_image_length );
323 We parse the metadata information here.
325 Although metadata information does not have to come in a fixed field
326 order, every metadata field consists of a tag byte followed by
327 parameters. In all cases known, the size including tag byte is
328 exactly four bytes in length.
331 #define ExtractXY(x, y) x = ((p[1]&0x0f)<<6) + (p[2]>>2); \
332 y = ((p[2]&0x03)<<8) + p[3];
334 static void ParseMetaInfo( decoder_t *p_dec, block_t *p_spu )
336 /* Last packet in subtitle block. */
338 decoder_sys_t *p_sys = p_dec->p_sys;
339 uint8_t *p = p_spu->p_buffer + p_sys->metadata_offset;
340 uint8_t *p_end = p + p_sys->metadata_length;
342 for( ; p < p_end; p += 4 )
346 case 0x04: /* subtitle duration in 1/90000ths of a second */
347 p_sys->i_duration = (p[1]<<16) + (p[2]<<8) + p[3];
350 msg_Dbg( p_dec, "subtitle display duration %lu secs",
351 (long unsigned int)(p_sys->i_duration / 90000) );
353 p_sys->i_duration *= 100 / 9;
356 case 0x0c: /* unknown */
358 msg_Dbg( p_dec, "subtitle command unknown 0x%0x 0x%0x 0x%0x 0x%0x",
359 (int)p[0], (int)p[1], (int)p[2], (int)p[3] );
363 case 0x17: /* coordinates of subtitle upper left x, y position */
364 ExtractXY(p_sys->i_x_start, p_sys->i_y_start);
367 msg_Dbg( p_dec, "start position (%d,%d)",
368 p_sys->i_x_start, p_sys->i_y_start );
372 case 0x1f: /* coordinates of subtitle bottom right x, y position */
376 ExtractXY(lastx, lasty);
377 p_sys->i_width = lastx - p_sys->i_x_start + 1;
378 p_sys->i_height = lasty - p_sys->i_y_start + 1;
381 msg_Dbg( p_dec, "end position (%d,%d), w x h: %dx%d",
382 lastx, lasty, p_sys->i_width, p_sys->i_height );
392 uint8_t v = p[0] - 0x24;
395 /* Primary Palette */
396 msg_Dbg( p_dec, "primary palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
397 (int)v, (int)p[1], (int)p[2], (int)p[3] );
400 p_sys->p_palette[v][0] = p[1]; /* Y */
401 p_sys->p_palette[v][1] = p[3]; /* Cr / V */
402 p_sys->p_palette[v][2] = p[2]; /* Cb / U */
411 uint8_t v = p[0] - 0x2c;
414 msg_Dbg( p_dec,"highlight palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
415 (int)v, (int)p[1], (int)p[2], (int)p[3] );
418 /* Highlight Palette */
419 p_sys->p_palette_highlight[v][0] = p[1]; /* Y */
420 p_sys->p_palette_highlight[v][1] = p[3]; /* Cr / V */
421 p_sys->p_palette_highlight[v][2] = p[2]; /* Cb / U */
426 /* transparency for primary palette */
427 p_sys->p_palette[0][3] = (p[3] & 0x0f) << 4;
428 p_sys->p_palette[1][3] = (p[3] >> 4) << 4;
429 p_sys->p_palette[2][3] = (p[2] & 0x0f) << 4;
430 p_sys->p_palette[3][3] = (p[2] >> 4) << 4;
433 msg_Dbg( p_dec, "transparency for primary palette 0..3: "
434 "0x%0x 0x%0x 0x%0x 0x%0x",
435 (int)p_sys->p_palette[0][3], (int)p_sys->p_palette[1][3],
436 (int)p_sys->p_palette[2][3], (int)p_sys->p_palette[3][3]);
441 /* transparency for highlight palette */
442 p_sys->p_palette_highlight[0][3] = (p[2] & 0x0f) << 4;
443 p_sys->p_palette_highlight[1][3] = (p[2] >> 4) << 4;
444 p_sys->p_palette_highlight[2][3] = (p[1] & 0x0f) << 4;
445 p_sys->p_palette_highlight[3][3] = (p[1] >> 4) << 4;
448 msg_Dbg( p_dec, "transparency for highlight palette 0..3: "
449 "0x%0x 0x%0x 0x%0x 0x%0x",
450 (int)p_sys->p_palette_highlight[0][3],
451 (int)p_sys->p_palette_highlight[1][3],
452 (int)p_sys->p_palette_highlight[2][3],
453 (int)p_sys->p_palette_highlight[3][3] );
458 /* offset to start of even rows of interlaced image, we correct
459 * to make it relative to i_image_offset (usually 4) */
460 p_sys->first_field_offset =
461 (p[2] << 8) + p[3] - p_sys->i_image_offset;
463 msg_Dbg( p_dec, "1st_field_offset %zu",
464 p_sys->first_field_offset );
469 /* offset to start of odd rows of interlaced image, we correct
470 * to make it relative to i_image_offset (usually 4) */
471 p_sys->second_field_offset =
472 (p[2] << 8) + p[3] - p_sys->i_image_offset;
474 msg_Dbg( p_dec, "2nd_field_offset %zu",
475 p_sys->second_field_offset);
481 msg_Warn( p_dec, "unknown sequence in control header "
482 "0x%0x 0x%0x 0x%0x 0x%0x", p[0], p[1], p[2], p[3]);
488 /*****************************************************************************
489 * DecodePacket: parse and decode an SPU packet
490 *****************************************************************************
491 * This function parses and decodes an SPU packet and, if valid, returns a
493 *****************************************************************************/
494 static subpicture_t *DecodePacket( decoder_t *p_dec, block_t *p_data )
496 decoder_sys_t *p_sys = p_dec->p_sys;
498 subpicture_region_t *p_region;
500 video_palette_t palette;
503 /* Allocate the subpicture internal data. */
504 p_spu = decoder_NewSubpicture( p_dec );
505 if( !p_spu ) return NULL;
507 p_spu->i_start = p_data->i_pts;
508 p_spu->i_stop = p_data->i_pts + p_sys->i_duration;
509 p_spu->b_ephemer = true;
511 /* Create new SPU region */
512 memset( &fmt, 0, sizeof(video_format_t) );
513 fmt.i_chroma = VLC_CODEC_YUVP;
514 fmt.i_aspect = VOUT_ASPECT_FACTOR;
515 fmt.i_width = fmt.i_visible_width = p_sys->i_width;
516 fmt.i_height = fmt.i_visible_height = p_sys->i_height;
517 fmt.i_x_offset = fmt.i_y_offset = 0;
518 fmt.p_palette = &palette;
519 fmt.p_palette->i_entries = 4;
520 for( i = 0; i < fmt.p_palette->i_entries; i++ )
522 fmt.p_palette->palette[i][0] = p_sys->p_palette[i][0];
523 fmt.p_palette->palette[i][1] = p_sys->p_palette[i][1];
524 fmt.p_palette->palette[i][2] = p_sys->p_palette[i][2];
525 fmt.p_palette->palette[i][3] = p_sys->p_palette[i][3];
528 p_region = subpicture_region_New( &fmt );
531 msg_Err( p_dec, "cannot allocate SPU region" );
532 decoder_DeleteSubpicture( p_dec, p_spu );
536 p_spu->p_region = p_region;
537 p_region->i_x = p_sys->i_x_start;
538 p_region->i_x = p_region->i_x * 3 / 4; /* FIXME: use aspect ratio for x? */
539 p_region->i_y = p_sys->i_y_start;
541 RenderImage( p_dec, p_data, p_region );
546 /*****************************************************************************
547 * ParseImage: parse and render the image part of the subtitle
548 *****************************************************************************
549 This part parses the subtitle graphical data and renders it.
551 Image data comes interlaced and is run-length encoded (RLE). Each
552 field is a four-bit nibbles that is further subdivided in a two-bit
553 repeat count and a two-bit color number - up to three pixels can be
554 described in four bits. What a 0 repeat count means is unknown. It
555 might be used for RLE extension. There is a special case of a 0
556 repeat count though. When the full nibble is zero, the rest of the
557 line is filled with the color value in the next nibble. It is
558 unknown what happens if the color value is greater than three. The
559 rest seems to use a 4-entries palette. It is not impossible that the
560 fill-line complete case above is not as described and the zero repeat
561 count means fill line. The sample code never produces this, so it
564 However we'll transform this so that that the RLE is expanded and
565 interlacing will also be removed. On output each pixel entry will by
566 a 4-bit alpha (filling 8 bits), and 8-bit y, u, and v entry.
568 *****************************************************************************/
569 static void RenderImage( decoder_t *p_dec, block_t *p_data,
570 subpicture_region_t *p_region )
572 decoder_sys_t *p_sys = p_dec->p_sys;
573 uint8_t *p_dest = p_region->p_picture->Y_PIXELS;
574 int i_field; /* The subtitles are interlaced */
575 int i_row, i_column; /* scanline row/column number */
576 uint8_t i_color, i_count;
579 bs_init( &bs, p_data->p_buffer + p_sys->i_image_offset,
580 p_data->i_buffer - p_sys->i_image_offset );
582 for( i_field = 0; i_field < 2; i_field++ )
584 for( i_row = i_field; i_row < p_sys->i_height; i_row += 2 )
586 for( i_column = 0; i_column < p_sys->i_width; i_column++ )
588 uint8_t i_val = bs_read( &bs, 4 );
592 /* Fill the rest of the line with next color */
593 i_color = bs_read( &bs, 4 );
595 memset( &p_dest[i_row * p_region->p_picture->Y_PITCH +
597 p_sys->i_width - i_column );
598 i_column = p_sys->i_width;
603 /* Normal case: get color and repeat count */
604 i_count = (i_val >> 2);
605 i_color = i_val & 0x3;
607 i_count = __MIN( i_count, p_sys->i_width - i_column );
609 memset( &p_dest[i_row * p_region->p_picture->Y_PITCH +
610 i_column], i_color, i_count );
611 i_column += i_count - 1;